AI Safety Fundamentals: Alignment

83 Avsnitt

1. Constitutional AI Harmlessness from AI Feedback

   Publicerades: 2024-07-19
2. Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback

   Publicerades: 2024-07-19
3. Illustrating Reinforcement Learning from Human Feedback (RLHF)

   Publicerades: 2024-07-19
4. Chinchilla’s Wild Implications

   Publicerades: 2024-06-17
5. Deep Double Descent

   Publicerades: 2024-06-17
6. Intro to Brain-Like-AGI Safety

   Publicerades: 2024-06-17
7. Eliciting Latent Knowledge

   Publicerades: 2024-06-17
8. Toy Models of Superposition

   Publicerades: 2024-06-17
9. Least-To-Most Prompting Enables Complex Reasoning in Large Language Models

   Publicerades: 2024-06-17
10. Discovering Latent Knowledge in Language Models Without Supervision

    Publicerades: 2024-06-17
11. ABS: Scanning Neural Networks for Back-Doors by Artificial Brain Stimulation

    Publicerades: 2024-06-17
12. Two-Turn Debate Doesn’t Help Humans Answer Hard Reading Comprehension Questions

    Publicerades: 2024-06-17
13. Imitative Generalisation (AKA ‘Learning the Prior’)

    Publicerades: 2024-06-17
14. An Investigation of Model-Free Planning

    Publicerades: 2024-06-17
15. Low-Stakes Alignment

    Publicerades: 2024-06-17
16. Gradient Hacking: Definitions and Examples

    Publicerades: 2024-06-17
17. Empirical Findings Generalize Surprisingly Far

    Publicerades: 2024-06-17
18. Compute Trends Across Three Eras of Machine Learning

    Publicerades: 2024-06-13
19. Worst-Case Thinking in AI Alignment

    Publicerades: 2024-05-29
20. Public by Default: How We Manage Information Visibility at Get on Board

    Publicerades: 2024-05-12